Matter Bounce in Hořava-Lifshitz Cosmology

Robert Brandenberger1,2
1) Department of Physics, McGill University, Montréal, QC, H3A 2T8, Canada and
2) Theory Division, CERN, CH-1211 Geneva, Switzerland
(Dated: April 22, 2009)
Hořava-Lifshitz gravity, a recent proposal for a UV-complete renormalizable gravity theory, may
lead to a bouncing cosmology. In this note we argue that Hořava-Lifshitz cosmology may yield a
concrete realization of the matter bounce scenario, and thus give rise to an alternative to inflation
arXiv:0904.2835v2 [hep-th] 22 Apr 2009

for producing a scale-invariant spectrum of cosmological perturbations. In this scenario, quantum

vacuum fluctuations exit the Hubble radius in the pre-bounce phase and the spectrum is transformed
into a scale-invariant one on super-Hubble scales before the bounce because the long wavelength
modes undergo squeezing of their wave-functions for a longer period of time than shorter wavelength
modes. The scale-invariance of the spectrum of curvature fluctuations is preserved during and after
the bounce. A distinctive prediction of this scenario is the amplitude and shape of the bispectrum.

PACS numbers: 98.80.Cq

I. INTRODUCTION down, as suggested more generally in the context of the

Recently, Hořava (based on the pioneering work of [1]
has proposed [2, 3] a model for quantum gravity which
is power-counting renormalizable and hence potentially
ultra-violet (UV) complete. This model does not have
the complete diffeomorphism invariance of General Rel-
ativity, but the action has a fixed point in the infrared
(IR) which is that of General Relativity with a negative
cosmological constant. In the UV, however, the theory
flows to a different fixed point, a fixed point at which
space and time scale differently and which has much bet-
ter UV behavior of perturbation theory. Since Hořava’s
theory is modelled after a scalar field model studied by
Lifshitz [4] in which the full Lorentz symmetry also only
emerges at an IR fixed point, the theory is now called
Hořava-Lifshitz gravity.
Specific solutions of the simplest version of Hořava-
Lifshitz gravity have recently been analyzed. In [5], ho-
mogeneous vacuum solutions with gravitational waves
were studied. in [6, 7], cosmological solutions with mat-
ter were explored, and in [8], black hole solutions were
analyzed, As pointed out in [6, 7], the analogs of the
Friedmann equations in Hořava-Lifshitz gravity include
a term which scales as dark radiation and contributes a
negative term to the energy density. Thus, it is possi-
ble in principle to obtain a nonsingular cosmological evo-
lution with the Big Bang of Standard and Inflationary
Cosmology replaced by a bounce.
In [2, 7], it was argued that the different ultraviolet
behavior of the theory might provide an alternative to
cosmological inflation for solving the problems of Stan-
dard Cosmology such as the horizon and flatness prob-
lems. Specifically, the divergence of the speed of light in
the far ultraviolet leads to the possibility of solving the
horizon problem as proposed a while back in [9, 10]. In
[6] it was emphasized that if the wavelength of fluctua-
tions penetrates the UV region, then the usual arguments
for the origin of a scale-invariant spectrum of cosmologi-
cal perturbations from an inflationary phase might break
“trans-Planckian problem” for inflationary fluctuations
[11, 12].
However, if Hořava-Lifshitz cosmology leads to a cos-
mological bounce, then it is not necessary to invoke a
period of inflationary expansion to produce the observed
spectrum of cosmological perturbations. The purpose of
this Note is to point out that Hořava-Lifshitz cosmol-
ogy may provide a UV-complete realization of the “mat-
ter bounce” scenario (see [13] for an introduction to this
scenario), an alternative to cosmological inflation for ex-
plaining the origin of the observed structure in the Uni-
verse.
As realized in [14, 15, 16], perturbations which start
out as quantum vacuum fluctuations and exit the Hubble
radius during a matter-dominated phase of contraction
acquire a scale-invariant spectrum. Given a non-singular
bouncing background cosmology, the fluctuations can be
followed unambiguously through the bounce. If the en-
ergy density at which the bounce occurs is smaller than
the Planck scale, then the wavelength of the fluctuations
which are being probed in today’s observations are in the
far IR (they are a fraction of a mm). Hence, the equa-
tions which describe these fluctuations are those of the IR
fixed point of the theory, which is the Einstein action in
the case of Hořava-Lifshitz cosmology. It has been shown
[17, 18, 19, 20] that, provided the duration of the bounce
phase is short compared to the wavelength of the fluctu-
ations being considered, then the spectrum of curvature
fluctuations is not changed during the bounce. Thus, a
scale-invariant spectrum of curvature perturbations will
persist in the post-bounce expanding phase. Specific pre-
dictions of the matter bounce scenario include a specific
form on the non-Gaussianities as measured by the am-
plitude and shape of the bispectrum [21].
In this Note we give an overview of the Hořava-Lifshitz
matter bounce cosmology, leaving details for future inves-
tigations. We begin with a space-time sketch depicting
the relevant phases (Figure 1). The vertical axis is time,
with t = 0 denoting the bounce time. In some early phase
 2

of contraction, the equation of state of matter is assumed

to be dominated by non-relativistic pressure-less matter,
in the same sense that our current expanding universe
is. For times between −tm and the bounce, the equa-
tion of state can be different from that of pressure-less
matter. The horizontal axis denotes co-moving spatial
coordinates. Vertical lines correspond to fixed co-moving
wavelengths, the dashed line is the co-moving Hubble ra-
dius H −1 . Fluctuations which cross the Hubble radius
during the matter phase of contraction acquire a scale-
invariant spectrum, those which cross later have a non-
trivial spectral slope whose magnitude depends on the
specific equation of state (see e.g. [22]).
The outline of this Note is as follows. We first re-
view the action of Hořava-Lifshitz gravity. Next, we re-
view the equations for cosmological solutions and study
the possibility of obtaining a bouncing cosmology with
a matter-dominated phase of contraction. In Section 4
we review the evolution of fluctuations in the contract-
ing phase of the matter bounce scenario, and in Section
5 we study how fluctuations pass through the bounce in
Hořava-Lifshitz cosmology. We conclude with a discus-
sion of some of the many open issues.

II. REVIEW OF HOŘAVA-LIFSHITZ GRAVITY

We begin with a brief review of Hořava-Lifshitz gravity FIG. 1: A space-time sketch of the matter bounce scenario.
[37] The dynamical variables are the lapse and shift func- The vertical axis is time, with t = 0 being the bounce time.
tions N and Ni , respectively, and the spatial metric gij The horizontal axis denotes co-moving distance. The curve
(roman letters indicate spatial indices) In terms of these with the label H is the Hubble radius H −1 - in co-moving
fields, the full metric is coordinates, the vertical line labeled by k denotes the co-
moving wavelength of a fluctuation mode. This mode crosses
ds2 = −N 2 dt2 + gij dxi + N i dt dxj + N j dt , (1)



the Hubble radius in the contracting phase before the time
−tc , when the period of matter-domination ends.
where the indices of N are raised and lowered using the
spatial metric gij .
The scaling symmetry of the coordinates in the sim-
The tensor ǫijk is the totally antisymmetric unit tensor,
plest version of Hořava-Liftshitz gravity is (we are fol-
λ is a dimensionless constant and Λ is related to the
lowing the notation of [7])
cosmological constant in the IR limit. The variables κ,
t → l3 t and xi → lxi . (2) w and µ are constants with mass dimensions −1, 0 and
1, respectively.
The full action of this version of Hořava-Lifshitz grav- In the IR limit, the action reduces to
ity is
√ 
Z
√ 2
Z
dtd3 x gN α(Kij K ij − λK 2 ) + ξR + σ , (6)

S = dtd3 x gN 2 (Kij K ij − λK 2 ) SE =
κ
κ2 κ2 µ ǫijk
− 4 Cij C ij + √ Ril ∇j Rkl (3) with
2w 2w2 g
2
κ2 µ2 κ2 µ2 1 − 4λ 2 α = (7)
Rij Rij + R + ΛR − 3Λ2 , κ2


−
8 8(1 − 3λ) 4 κ2 µ2
ξ = Λ and (8)
where 8(1 − 3λ)
1
κ2 µ2
Kij = g˙ij − ∇i Nj − ∇j Ni , (4) σ = −3 Λ2 . (9)
2N 8(1 − 3λ)
and Cij is the Cotton tensor
In order to obtain the Einstein action, we require λ = 1.
ǫijk 1 In this case, the variables of the Hořava-Lifshitz action go
ij
= √ ∇k Rij − Rδij .


C (5)
g 4 over into the following expressions for the speed of light
 3

c, Newton’s gravitational constant G and the effective The key new term in the cosmological equations of mo-
cosmological constant ΛE : tion is the second to last term on the right-hand sides of
r (15) and (16). This term corresponds to “dark radiation”
ξ with a negative energy density. This term is present only
c = , (10)
α if the spatial curvature of the metric is non-vanishing. If
the energy density of regular matter increases less fast
r
ξ
16πG = , (11) than a−4 as the scale factor decreases, the dynamics will
α3
σ lead to a cosmological bounce provided that [38]
ΛE = − , (12)
2α ρ
( − p) > 0 . (20)
In the following we will consider scalar field matter in 12
the contracting phase. The action for matter is To obtain a bounce in Hořava-Lifshitz cosmology we
will assume that matter in the pre-bounce epoch is de-
√
Z
SM = dtd3 x gN Lm , (13) scribed by a scalar field ϕ with a potential
1 2 2
where the matter Lagrangian Lm depends on the scalar V (ϕ) = m ϕ . (21)
2
matter field ϕ and the metric. In the IR limit, this action
reduces to the usual scalar field matter action in curved As studied in detail in quintom bounce [19, 20, 34] sce-
space-time. The form of the scalar field Lagrangian valid narios, we take the scalar field to be oscillating during
also in the UV is given in [7] but will not be used in this the contracting phase, with an amplitude A(t).
Note. As the universe contracts, the amplitude A(t) ∼
a(t)−3/2 will increase. Once the amplitude reaches the
value
III. MATTER BOUNCE IN HOŘAVA-LIFSHITZ
COSMOLOGY Acrit = (12π)−1/2 mpl , (22)
where mpl is the Planck mass, then the field oscillations
To obtain the equations for Hořava-Lifshitz cosmology will stop and ϕ will enter a “slow climb” phase, the time
we assume that the metric is homogeneous and isotropic, reversal of the inflationary slow-roll phase. During this
i.e. phase, the matter energy density is approximately con-
N = N (t), Ni = 0 and gij = a2 (t)γij , (14) stant but the scale factor is rapidly decreasing. Hence,
the dark radiation term in the Hubble equation rapidly
where γij is a maximally symmetric constant curvature catches up with the matter energy term. Since in the
metric. We will denote the spatial curvature parameter slow climb phase the pressure of matter is negative, the
by k̄. condition (20) is satisfied. Thus, we obtain a cosmologi-
The equations of motion are obtained by varying the cal bounce.
action with respect to N , a, and ϕ, and setting N = 1 at Note that the “slow climb” phase is unstable with re-
the end of the calculation. The resulting equations are spect to the presence of the second mode in the scalar
field equation of motion, a mode which is exponentially
k̄ ΛE 2k̄ 2 (ζ + 3η) ρ decaying in an inflationary slow-roll phase and thus en-
H2 = − − − + ,(15)
a2 3 αa4 6α sures that the slow-roll trajectory in large-field inflation
3  k̄ ΛE k̄ 2 (ζ + 3η) p is a local attractor [35]. In the “slow climb” phase, the
Ḣ + H 2 = − 2 −

+ + (, 16)
2 2a 2 4αa 4 4α second mode is increasing. Thus, the “slow climb” tra-
jectory is a repeller [39]. However, coming out of the
and
oscillatory phase, the initial amplitude of the unstable
ϕ̈ + 3H ϕ̇ + V ′ = 0 , (17) mode is sufficiently small such that the instability does
not have time to develop before the bounce takes place.
where H = ȧ/a, p and ρ are the pressure and energy
density of the scalar matter field, respectively, a prime
denotes the derivative with respect to ϕ, the dimension- IV. FLUCTUATIONS IN THE MATTER
less constant η BOUNCE SCENARIO

κ2 µ2 (1 − 4λ) In the following we assume that the contracting phase

η = (18)
32(1 − 3λ) before the bounce was first dominated by cold matter,
matter with an equation of state w = 0, where w is the
is the coefficient of the R2 term in the gravitational ac-
ratio of pressure p divided by energy density ρ. We will
tion, and the dimensionless constant ζ is given by
now review how an initial vacuum spectrum of cosmolog-
κ2 µ2 ical perturbations on sub-Hubble scales in the contract-
ζ = − . (19) ing phase develops into a scale-invariant spectrum for
8
 4

wavelengths which exit the Hubble radius in the matter- R scales as η −3 , which is decaying in an expanding uni-
dominated phase. verse but growing in a contracting phase. It is this growth
As has proven to be convenient in inflationary cosmol- which is responsible for turning an initial vacuum spec-
ogy, we track the cosmological fluctuations in terms of trum of fluctuations into a scale-invariant one.
the variable R, the curvature fluctuations in co-moving To see this, let us compute the power spectrum of R
coordinates [23, 24, 25, 26] . This variable is conserved at on super-Hubble scales late in the contracting phase:
phase transitions and is constant on super-Hubble scales
in an expanding universe. PR (k, η) ∼ k 3 |vk (η)|2 a−2 (η) (30)
If we work in longitudinal gauge in which the metric ηH (k) 2
∼ k 3 |vk (ηH (k))|2 ∼ k 3−1−2


in the absence of anisotropic stress takes the form η
∼ const .
ds2 = a2 (η) (1 + 2Φ)dη 2 − (1 − 2Φ)dx2 ,
 
(23)
In this first step, we have used the definition of the power
where η is conformal time, x are co-moving spatial co- spectrum, replaced R by v via (25) and used the scaling
ordinates) and Φ(x, η) describes the metric fluctuations, z(η) ∼ a(η). In the second step, we made use of the
then R is given by (modulo terms which are suppressed growth of the c1 mode, the dominant mode, to relate the
on super-Hubble scales) value of v at late times to its value at the time ηH (k)
when the mode k crosses the Hubble radius. Finally, in
2 ′
1 the last step we insert the vacuum spectrum for v on sub-
R = HΦ + Φ + Φ, (24)
3 1+w Hubble scales and the Hubble radius crossing condition
ηH (k) ∼ k −1 .
H denoting the Hubble expansion rate in conformal time
and a prime indicating the derivative with respect of η.
The variable R is closely related to the variable v (see V. EVOLVING FLUCTUATIONS THROUGH
[27] for an in depth review of the theory of cosmologi- THE BOUNCE
cal fluctuations and [28] for an introductory overview) in
terms of which the action for cosmological fluctuations
If the bounce phase is short compared to the wave-
has canonical kinetic term:
length of the fluctuations which are being followed, then
v the spectrum of R is unchanged through the bounce.
R = (25)
z This result can be obtained by explicitly evolving fluctua-
tions through a non-singular bounce using the equations
where z is a function of the background which is pro-
of motion for fluctuations which follow from Einstein’s
portional to the scale factor a as long as the equation of
theory (see e.g. [17, 18, 19, 20]). This result also agrees
state of matter in constant.
with what is obtained by replacing the bounce phase by a
The equation of motion for the Fourier mode vk of v is
matching surface and making use of the Hwang-Vishniac
′′
′′
z
[29] (Durrer-Mukhanov [30]) matching conditions.
vk + k 2 − vk = 0 . (26) If the energy density at the bounce is of the order of
z
(1016 GeV)4 , the wavelength of a mode which corresponds
This shows that on length scales larger than the Hub- to the current Hubble radius is about 1mm, i.e. in the
ble radius, where the k 2 term is negligible, v does not far IR. Since in the Hořava-Lifshitz bounce, the bounce
oscillate, its time evolution being determined by the the time is set by the UV scale, it is well justified to assume
gravitational background, whereas on sub-Hubble scales that in the context of the use of the Einstein equations
vk is oscillating with approximately constant amplitude. for the gravitational fluctuations the spectrum of R does
On super-Hubble scales, the equation of motion for not change across the bounce.
vk in a universe which is undergoing matter-dominated In addition, again since the scales we are interested
contraction is in are in the far IR, it should be justified to use the
′′
IR limit of Hořava-Lifshitz cosmology to propagate the
vk = 2η −2 vk , (27) fluctuations [40]
Thus, we argue that the scale-invariance of the spec-
which has the general solution trum of cosmological perturbations will be scale-invariant
after the bounce. Since R is constant on super-Hubble
vk (η) = c1 η −1 + c2 η 2 , (28)
scales in the post-bounce expanding phase, it then imme-
where c1 and c2 are constants. Since for a matter domi- diately follows that the spectrum of fluctuations at late
nated phase times will be scale-invariant.
A bouncing cosmology in the context of Hořava-
a(η) ∼ η 2 (29) Lifshitz gravity can thus provide an alternative to infla-
tion for providing a scale-invariant spectrum of cosmo-
it follows that the c2 mode is the mode for which R is logical perturbations, provided that we begin in the con-
constant on super-Hubble scales, whereas for the c1 mode tracting phase with quantum vacuum fluctuations and

provided that the relevant scales exit the Hubble radius
in a period of cold matter domination (the case of initial
thermal fluctuations in analyzed in [33]).
Since the curvature perturbation R grows on super-
Hubble scales, a matter bounce leads to a larger ampli-
tude of non-Gaussianities than slow-roll single-field infla-
tion. Since it is a different mode of R which dominates,
the shape of the non-Gaussianities is also different from
what is obtained in slow-roll single-field inflation models.
The specific predictions for the amplitude and shape of
the three-point function (the “bispectrum”) were worked
out in [21]. In particular, the predicted amplitude of the
bispectrum is very close to the level which could be de-
tected using the Planck satellite experiment.
VI. CONCLUSIONS AND DISCUSSION
In this paper we have shown how to obtain a “matter
bounce” if Hořava-Lifshitz gravity. Such a bounce is ob-
tained because of a “dark radiation” term which appears
in the equations of motion for cosmological solutions, a
term which stems from the terms in the quantum grav-
ity theory which appear in the UV and help render the
theory renormalizable. Note, however, that the presence
of the dark radiation term requires non-vanishing spatial
curvature.
To obtain a cosmological bounce it is important that
no source of matter is present which redshifts equally fast
or faster than that of the dark radiation term. Only in
this case the energy density of dark radiation can grown
with respect to the regular matter energy, a condition
which is required to obtain a bounce. This condition
appears to be rather restrictive since it even rules out
regular radiation before the bounce.
We have presented a model in which a bounce can be
obtained. In this model, matter is modeled by a scalar
field with a standard mass term. The scalar field oscil-
lates at early times in the contracting phase, leading to
a matter equation of state which is that of cold matter.
Once the amplitude of scalar field oscillations reaches a
critical value, the field enters a deflationary slow-climb
phase during which its energy density is approximately
constant and its pressure is negative. Hence, a bounce
occurs.
In the matter bounce model thus constructed, initial
quantum vacuum fluctuations which exit the Hubble ra-
dius in the contracting matter-dominated phase acquire
[1] P. Horava, “Membranes at Quantum Criticality,” JHEP
0903, 020 (2009) [arXiv:0812.4287 [hep-th]].
[2] P. Hořava, “Quantum Gravity at a Lifshitz Point,” Phys.
Rev. D 79, 084008 (2009) [arXiv:0901.3775 [hep-th]].
[3] P. Hořava, “Spectral Dimension of the Universe in Quan-
tum Gravity at a Lifshitz Point,” Phys. Rev. Lett. 102,
161301 (2009) [arXiv:0902.3657 [hep-th]].
5
a scale-invariant spectrum of curvature fluctuations, as
already envisioned in [14, 15, 16] and recently studied in
detail in [20]. Thus, one of the main messages of this Note
is that it is not necessary to force a period of inflationary
expansion into Hořava-Lifshitz cosmology. The alterna-
tive matter bounce scenario predicts an amplitude of the
normalized bi-spectrum is the order of 1, and a specific
shape of this three-point function, as studied in detail in
[21]. These specific predictions are potentially within the
reach of upcoming CMB missions such as PLANCK.
To obtain a successful late-time cosmology, the model
presented here must be supplemented with a mechanism
to transfer the energy at late times to Standard Model
matter and radiation. If we include such matter in the ba-
sic Lagrangian, then an initial condition problem arises:
in order for a bounce to occur, the initial energy den-
sity of radiation must be so low that the radiation never
comes to dominate during the contracting phase. All of
these issues deserve further study.
While this Note was being prepared for submission,
a very interesting paper by Mukohyama [36] appeared
showing that in the UV region, fluctuations of a scalar
field in Hořava-Lifshitz gravity acquire a scale-invariant
spectrum, a spectrum which can be later transformed to
curvature fluctuations via a transfer from entropy to adi-
abatic modes such as the curvaton mechanism. Thus,
one can obtain a scale-invariant spectrum of cosmolog-
ical fluctuations. An advantage of this mechanism is
that it also operates in a background cosmology with-
out a bounce (e.g. in the case of zero spatial curvature).
However, in the case of a matter bounce background, it
is unclear whether the scaling of the correlation func-
tions used in [36] would extend to the large IR scales
required to match with observations, the scales on which
our mechanism works nicely.
Acknowledgments
The author wishes to thank G. Calcagni, P. Hořava
and S. Mukohyama for comments on the draft of this
Note. This work is supported by an NSERC Discovery
Grant and by the Canada Research Chairs Program. The
author wishes to acknowledge the hospitality of the In-
stitute of High Energy Physics in Beijing, where a lot of
the initial work on the matter bounce scenario was de-
veloped, and of the CERN Theory Division, where the
ideas presented here were worked out.
[4] E.M. Lifshitz, “On the Theory of Second-Order Phase
Transitions I & II”, Zh. Eksp. Teor. Fiz. 11, 255 & 269
(1941).
[5] T. Takahashi and J. Soda, “Chiral Primordial Gravi-
tational Waves from a Lifshitz Point,” arXiv:0904.0554
[hep-th].
[6] G. Calcagni, “Cosmology of the Lifshitz universe,”

arXiv:0904.0829 [hep-th].
[7] E. Kiritsis and G. Kofinas, “Hořava-Lifshitz Cosmology,”
arXiv:0904.1334 [hep-th].
[8] H. Lu, J. Mei and C. N. Pope, “Solutions to Hořava Grav-
ity,” arXiv:0904.1595 [hep-th].
[9] J. W. Moffat, “Superluminary Universe: A Possible So-
lution To The Initial Value Problem In Cosmology,” Int.
J. Mod. Phys. D 2, 351 (1993) [arXiv:gr-qc/9211020].
[10] A. J. Albrecht and J. Magueijo, “A time varying speed of
light as a solution to cosmological puzzles,” Phys. Rev.
D 59, 043516 (1999) [arXiv:astro-ph/9811018].
[11] R. H. Brandenberger, “Inflationary cosmology: Progress
and problems,” arXiv:hep-ph/9910410.
[12] J. Martin and R. H. Brandenberger, “The trans-
Planckian problem of inflationary cosmology,” Phys.
Rev. D 63, 123501 (2001) [arXiv:hep-th/0005209];
R. H. Brandenberger and J. Martin, “The robustness of
inflation to changes in super-Planck-scale physics,” Mod.
Phys. Lett. A 16, 999 (2001) [arXiv:astro-ph/0005432].
[13] R. H. Brandenberger, “Alternatives to Cosmological In-
flation,” arXiv:0902.4731 [hep-th].
[14] D. Wands, “Duality invariance of cosmological per-
turbation spectra,” Phys. Rev. D 60, 023507 (1999)
[arXiv:gr-qc/9809062].
[15] F. Finelli and R. Brandenberger, “On the generation of a
scale-invariant spectrum of adiabatic fluctuations in cos-
mological models with a contracting phase,” Phys. Rev.
D 65, 103522 (2002) [arXiv:hep-th/0112249].
[16] L. E. Allen and D. Wands, “Cosmological perturbations
through a simple bounce,” Phys. Rev. D 70, 063515
(2004) [arXiv:astro-ph/0404441].
[17] R. Brandenberger, H. Firouzjahi and O. Saremi, “Cos-
mological Perturbations on a Bouncing Brane,” JCAP
0711, 028 (2007) [arXiv:0707.4181 [hep-th]].
[18] S. Alexander, T. Biswas and R. H. Brandenberger, “On
the Transfer of Adiabatic Fluctuations through a Nonsin-
gular Cosmological Bounce,” arXiv:0707.4679 [hep-th].
[19] Y. F. Cai, T. Qiu, R. Brandenberger, Y. S. Piao and
X. Zhang, “On Perturbations of Quintom Bounce,”
JCAP 0803, 013 (2008) [arXiv:0711.2187 [hep-th]].
[20] Y. F. Cai, T. t. Qiu, R. Brandenberger and X. m. Zhang,
“A Nonsingular Cosmology with a Scale-Invariant Spec-
trum of Cosmological Perturbations from Lee-Wick The-
ory,” arXiv:0810.4677 [hep-th].
[21] Y. F. Cai, W. Xue, R. Brandenberger and X. Zhang,
“Non-Gaussianity in a Matter Bounce,” arXiv:0903.0631
[astro-ph.CO].
[22] H. Li, J. Q. Xia, R. Brandenberger and X. Zhang, “Con-
straints on Models with a Break in the Primordial Power
Spectrum,” arXiv:0903.3725 [astro-ph.CO].
[23] J. M. Bardeen, “Gauge Invariant Cosmological Pertur-
bations,” Phys. Rev. D 22, 1882 (1980).
[24] J. M. Bardeen, P. J. Steinhardt and M. S. Turner, “Spon-
6
taneous Creation Of Almost Scale - Free Density Pertur-
bations In An Inflationary Universe,” Phys. Rev. D 28,
679 (1983).
[25] R. H. Brandenberger and R. Kahn, “Cosmological Per-
turbations In Inflationary Universe Models,” Phys. Rev.
D 29, 2172 (1984).
[26] D. H. Lyth, “Large Scale Energy Density Perturbations
And Inflation,” Phys. Rev. D 31, 1792 (1985).
[27] V. F. Mukhanov, H. A. Feldman and R. H. Branden-
berger, “Theory of cosmological perturbations. Part 1.
Classical perturbations. Part 2. Quantum theory of per-
turbations. Part 3. Extensions,” Phys. Rept. 215, 203
(1992).
[28] R. H. Brandenberger, “Lectures on the theory of cosmo-
logical perturbations,” Lect. Notes Phys. 646, 127 (2004)
[arXiv:hep-th/0306071].
[29] J. c. Hwang and E. T. Vishniac, “Gauge-invariant joining
conditions for cosmological perturbations,” Astrophys. J.
382, 363 (1991).
[30] N. Deruelle and V. F. Mukhanov, “On matching condi-
tions for cosmological perturbations,” Phys. Rev. D 52,
5549 (1995) [arXiv:gr-qc/9503050].
[31] T. Biswas, R. Brandenberger, A. Mazumdar and
W. Siegel, “Non-perturbative gravity, Hagedorn
bounce and CMB,” JCAP 0712, 011 (2007)
[arXiv:hep-th/0610274].
[32] T. Biswas, A. Mazumdar and W. Siegel, “Bouncing uni-
verses in string-inspired gravity,” JCAP 0603, 009 (2006)
[arXiv:hep-th/0508194].
[33] Y. F. Cai, W. Xue, R. Brandenberger and X. Zhang,
“Thermal Fluctuations and Bouncing Cosmologies,”
arXiv:0903.4938 [hep-th].
[34] Y. F. Cai, T. Qiu, Y. S. Piao, M. Li and X. Zhang,
“Bouncing Universe with Quintom Matter,” JHEP 0710,
071 (2007) [arXiv:0704.1090 [gr-qc]].
[35] R. H. Brandenberger and J. H. Kung, “Chaotic Inflation
As An Attractor In Initial Condition Space,” Phys. Rev.
D 42, 1008 (1990).
[36] S. Mukohyama, “Scale-invariant cosmological perturba-
tions from Hořava-Lifshitz gravity without inflation,”
arXiv:0904.2190 [hep-th].
[37] This section is based completely on the analysis of [7].
[38] The contributions scaling like curvature or a cosmological
constant can be neglected.
[39] We thank Misao Sasaki and Takahiro Tanaka for discus-
sions on this point.
[40] This claim should be justified with an explicit calcula-
tion in the same way that the corresponding claim was
justified [31] in the bouncing scenario obtained by using
the special higher derivative gravitational action of [32]
which is ghost-free about Minkowski space-time.